Recognition of arm activities based on Hidden Markov Models for natural interaction with service robots

This research presents a novel way of representing human motion and recognizing human activities from the skeleton output computed from RGB-D data from vision-based motion capture systems. The method uses a representation of the skeleton which is invariant to rotation and translation, based on Orthogonal Direction Change Chain Codes, as observations for a single Discrete Connected Hidden Markov Model formed by a set of multiple Hidden Markov Models for simple activities, which are merged using a grammar-based structure. The purpose of this research is to provide a service robot with the capability of human activity awareness, which can be used for action planning with implicit and indirect Human-Robot Interaction.